College of Exploration · Est. 1991
Introducing

Confabula

A platform for asynchronous collaborative intelligence —
where communities build knowledge together across time,
and the conversation outlasts the meeting.

Confabula is not a forum. Not a chat tool. It is fundamentally a human-to-human space — structured discussion, threaded over time, where communities think together without AI involvement of any kind. For communities that want to go further, Augur is an optional addition: AI that reads the pattern of collective reasoning and surfaces what the group is becoming.

"Knowledge is not stored.
It is performed —
through conversation,
across time." — The Confabula approach

Most platforms treat knowledge as content to be delivered and consumed. Confabula treats it as something that happens between people — built, contested, revised, and deepened through genuine exchange.

This is not incidental. It is the foundational commitment of the platform, inherited from fifty years of practice in collaborative online learning and grounded in Gordon Pask's Conversation Theory: a concept is only understood when a learner can explain it to another.

Time-shifted dialogue lets participants think carefully, consult sources, and change their minds. The graph that accumulates from this exchange becomes the community's navigable intellectual history.

Structured for sustained inquiry

1
Interfaces & Conferences

Each workshop, course, or community gets its own branded interface with a custom URL, logo, and CSS. Conferences group related discussions. Access levels from read-only to organiser give precise control over participation.

2
Items & Threaded Responses

Members raise items for discussion. Others respond. Threads develop over days and weeks. The model is inherited from fifty years of collaborative conferencing — it works because it respects how people actually think.

3
The Knowledge Garden

Submit a YouTube video, article, or document. The pipeline extracts concepts and relationships into the knowledge graph, then seeds the conference with conversation starters — not summaries. Learn more →

4
Confabula Augur

On key questions, members register positions and confidence as they reason. AI reads the structure of the thread, tracks how views shift, and names the territory the group has not yet entered. Learn more →

5
The Living Graph

Every source, concept, position shift, and influence edge becomes a node or relationship in Neo4j. Over time the graph is the community's collective memory — navigable, queryable, and growing. The intellectual lineage of every idea is always traceable to its origin.

Fifty years of collaborative computing

1975

Confer

Mainframe conferencing software from the University of Michigan. The original model of conferences, items, and threaded responses — still the structural core of Confabula today.

1990s

Caucus

Screen Porch LLC extended Confer into a full web platform with CML templating, the interface system, and access control. Widely deployed in higher education.

1991 –

College of Exploration

Peter Tuddenham and Dr. Kristina Bishop built the College on Caucus. Over 18,000 educators reached through GOSE workshops. Foundational influence on national ocean literacy policy.

2026

Confabula

Complete rewrite on Python / FastAPI / Neo4j. Every core concept preserved. Fifty years of practice, rebuilt graph-native — with Augur and the Knowledge Garden as new growth.

Built for sustained intellectual work

Discussion

Conferences, Items & Responses

The proven Caucus model: threaded discussions that develop over weeks. Every response is a turn in a conversation, not a broadcast.

Interface System

Branded Workspaces

Each interface is a distinct environment — its own URL, logo, CSS, and conference set. Run multiple programmes from one platform without overlap.

Access Control

Five Permission Levels

From read-only to organiser, with granular per-conference control. Open enrolment, self-registration, or invite-only — configurable per interface.

Knowledge Graph

Neo4j Graph Database

Every concept, source, item, and participant relationship is a graph node or edge. The knowledge structure is first-class — navigable, queryable, and cumulative.

Provenance

Every Idea Traceable

SEEDED_FROM edges link every discussion item back to its originating source. The intellectual lineage of a community's ideas is always recoverable.

Infrastructure

Robust & Open

Python / FastAPI / Nginx / Neo4j. Daily automated backups. Self-hosted or Hetzner cloud. No vendor lock-in. Built by the College of Exploration — in use since 1991.

Two tools that extend the commons

New · Intelligence
⚖️

Confabula Augur

AI-augmented collective deliberation. Members register positions and confidence as they reason. Augur reads the structure of the thread, tracks how views shift over time, synthesises arguments against the knowledge graph, and names the territory the group has not yet entered. The asynchronous equivalent of real-time swarm intelligence — with trajectory, not just destination.

What is Augur
New · Ingest
🌱

The Knowledge Garden

Submit a YouTube video with a short framing note. The pipeline fetches the transcript, extracts concepts and relationships into the knowledge graph, and seeds your conference with conversation starters — not a summary, but questions that open territory. Over time, shared concepts across multiple videos converge to the same graph node. The garden becomes a commons.

About the Garden

Communities that need to think, not just communicate

🎓

Educators & Facilitators

Run sustained online programmes where participants engage deeply with material over weeks. Each cohort gets its own branded interface. The Garden turns your reading list into structured dialogue.

🔬

Research Communities

Build a shared knowledge graph from your field's sources. Track how your community's understanding evolves. Augur surfaces the questions you haven't answered yet.

🌐

Communities of Practice

The original Caucus vision: practitioners sharing observations, building shared vocabulary, developing collective expertise — in continuous asynchronous conversation with AI that reads the pattern.